Comparison of Propranolol
and Hydrochlorothiazide for the
Initial Treatment of Hypertension

II. Results of Long-term Therapy

Veterans Administration Cooperative Study Group on Antihypertensive Agents

o As described in the preceding communication, either propranolol
hydrochloride or hydrochlorothiazide were randomly allocated in a double-
blind manner to 683 patlents with Initial diastolic BP in the range of 95 to 114
mm Hg. Of this number, 394 entered the long-term treatment phase. During
the subsequent 12 months of long-term treatment, hydrochlorothiazlde was
more effective than propranolol in controlling BP (mean reductions, -17.5
-13.1 mm Hg with hydrochlorothiazide compared with -8.31-l 1.3 with
propranolol. After treatment with hydrochlorothiarlde, a greater percentage
of patients achieved the goal diastolic BP of less than 90 mm Hg (65.5%
compared with 52.8% taking propranolol). Also during treatment, fewer
patients receiving hydrochlorothiaride required termination as compared
with those receiving propranolol; comparative dosage requirements were
lower; additional, titration during long-term treatment was required less
often, and BP remained lower after withdrawal of the active drugs. However,
biochemical abnormalities were greater with hydrochlorothiazide. Although
not statistically significant, the antihypertensive effects of hydrochlorothia-
zide were greater in blacks than in whites. Whites, on the other hand, had a
greater response to propranolol than blacks, although it was still less than
the response of the whites to hydrochlorothiazide.

(JAMA 1982;248:2004-2011)

THIS study was designed to deter-
mine the degree of BP lowering that
can be achieved with propranolol
hydrochloride as compared with hy-
drochlorothiazide and to examine the
dosage distribution of each drug
required to achieve control of BP.
These questions have clinical rele-
vance because of recommendations
that propranolol replace thiazide di-
uretics as the primary treatment for
hypertension." This recommendation
involves mostly theoretical considera-
tions based either on renin profiling','

From the Cooperative Studies Program, Medical
Research Service of the Veterans Administration,
Washington, DC.

Reprint requests to the Veterans Administration
Medical Center, 50 Irving St NW, Washington, DC
20422 (Edward D. Freis, MD).

or on supposedly adverse biochemical
effects of thiazides as compared with
the possibly beneficial antiarrhyth-
mic effects of the &blockers.'

In a prior cooperative study carried
out by our group, propranolol alone
controlled the BP in 52% of the
patients with mild hypertension.'

See also p 1996.

However, no comparison was made
with diuretic alone. The present study
is designed to compare the two single
entities. The short-term responses to
these two agents are presented sepa-
rately in THE JOURNAL (~1996). The
present report is concerned with a
comparison of the long-term effec-
tiveness of propranolol and hydro-

chlorothiazide during 12 months of
continuous treatment. Data also are
presented concerning BP and other
. .-
indices before treatment and during a
final placebo period.

SUBJECTS AND METHODS

The design of the prerandomization and
early postrandomization (titration) period
of the study are described in the preceding
communication. Briefly, this consisted of a
prerandomization placebo period of four
weeks' duration to obtain baseline data
and to determine eligibility defined as an
average diastolic BP at two consecutive
visits in the range of 95 to 114 mm Hg and
pill counts of the placebo within a desig-
nated acceptable range with respect to
compliance. Consenting, eligible patients
were randomly assigned in a double-blind
fashion to either propranolol hydrochlo-
ride, 80 mg daily, or hydrochlorothiazide,
50 mg daily, both given in equal twice-
daily divided doses. Doses of propranolol
hydrochloride were titrated to as high as
640 mg per day, or until the diastolic BP
fell below 90 mm Hg or there were side
effects. Doses of hydrochlorothiazide were
titrated similarly up to a level of 200 mg
daily. A maximum of seven clinic visits
one to two weeks apart were permitted to
complete the titration of dosage. To enter
the long-term treatment phase of the
study the patient had to achieve an aver-
age diastolic BP during the last two con-
secutive visits of the acute titration phase
below 100 mm Hg and at least 6 mm Hg
less than the baseline average. The long-
term treatment phase consisted of 48
weeks of continuous treatment.

Additional titration of dosage was per-
mitted during the long-term treatment
period if all of the following conditions
were present: (1) diastolic BP was 90 mm
Hg or higher; (2) patient had been com-

2004  JAMA, Ott 22129, 1982-Vol 248, No. 16          Propranolol and Hydrochlorothiazide, Part II-VA Study Group


pliant as evidenced by pill counts (return
of 20% or less of prescribed number of
tablets); (3) dose level had been changed
less than three months previously; (4)
patient had not yet passed the ninth week
of the long-term treatment period; (5)
there were no severe side effects; and (6)
the maximum permitted dose level had not
yet been reached. If the dose had been
titrated without producing additional an-
tihypertensive effects, hydrochlorothia-
zide administration was dropped to the
level where no further fall had occurred
with subsequent increases of dose. Also, if
severe dose-related side effects or serious
hypotensive symptoms developed, the dose
could be reduced to the next lower level
after consulting with the chairman of the
study.

Patients were terminated from the
study if the diastolic BP was greater than
119 mm Hg on any single visit. Patients
with diastolic BP between 105 and 119 mm
Hg who had been previously controlled or
had finished their initial dosage titration
were seen again in one week. If the
diastolic BP was still greater than 104 mm
Hg the patient was terminated from the
study after dose tapering. Patients with
two successive visits at which diastolic BP
was greater than the pretreatment aver-
age also were terminated.

The 48-week interval during the long-
term treatment period was subdivided into
12 visits at intervals of four weeks. This
was followed by two weekly visits for dose
tapering and two further weekly visits for
the final placebo period. Tapering was
carried out using blister packs containing
both active drug and placebo tablets so
that the patient was not aware of when his
dosage was being reduced. Tapering was
carried out to protect the patient against
potential complications that may result
from the sudden discontinuation of a fl-
blocker.

Laboratory studies, including the stan-
dard urinalysis, complete blood cell count,
and blood chemistries determined by auto-
mated analyzers, were carried out just
before the beginning of the dosage titra-
tion phase, after completing dosage titra-
tion, midway and at the end of the 12-
month long-term treatment phase, and at
the end of the two weeks final placebo
phase. The ECG and x-ray film of the
chest also were taken at the beginning and
end of the long-term treatment phase. The
ECG was repeated at the end of the final
placebo phase. Special tests included plas-
ma renin activity studies, which were
carried out in five hospitals, and blood
glucose tolerance testing determined in
two hospitals. The results of these tests
will be reported separately.

A sample size of 300 patients random-
ized to each regimen, propranolol or
hydrochlorothiazide, was considered suffi-
cient to detect a 20% difference in the

proportion of patients successfully com-
pleting the long-term treatment period.
The criteria for effectiveness were the
absolute number and percent of patients
entering the long-term treatment phase
who in the seated position achieved a
diastolic BP of 90 mm Hg or less at the end
of the long-term treatment period. Stu-
dent's t test, x1 analysis of variance
(ANOVA), and regression analyses were
used to assess statistically significant dif-
ferences (P<.O5) between groups of data.

The mean age of the patients entering
the long-term treatment phase was 49.2
years for the propranolol group and 50.2
years for the patients receiving hydrochlo-
rothiazide. The racial distribution was
52% blacks and 48% whites. There was no
significant difference in the black-white
ratio of patients receiving propranolol
compared   with  hydrochlorothiazide-
treated patients. The percentage of pa-
tients receiving hydrochlorothiazide was
slightly greater, being 53.8% as compared
with 46.2% taking propranolol. This was
due to the greater number of terminations
in the group receiving propranolol during
the earlier titration phase of the study.

RESULTS

Of 906 patients who entered the
prerandomization placebo phase of
the study, 683 were randomized. How-
ever, only 491 of these entered the
study early enough to be eligible for
long-term treatment. Of this number,
394 completed the dose titration
phase of the study and met the crite-
ria for entering the long-term treat-
ment phase. Among the 394 patients
who entered the long-term treatment
phase, 302 completed the 12 month
follow-up, while 92 were terminated
for the reasons described herein.

Changes in BP

The pretreatment baseline diastolic
BP averaged 101.5 mm Hg in patients
who entered the long-term treatment
phase and was not significantly dif-
ferent in the two drug groups. The
changes in BP from the prerandomi-
zation baseline period in the 302
patients who completed the long-term
treatment period averaged -8.3/
-11.3 mm Hg with propranolol and
-17.5/-13.1 mm Hg with hydrochlo-
rothiazide. The reductions were sig-
nificantly greater with hydrochloro-
thiazide than with propranolol for
both the systolic (P<.OOl) and dia-
stolic (P<.OOl) BP.

The average diastolic BP at the end
of the preceding titration period was
less than the average at the end of

long-term treatment in both treat-
ment groups, the rise being greater
with propranolol than with hydro-
chlorothiazide. The mean elevations
were +7.0/+3.5 mm Hg with propran-
0101 and +1.8/+1.0 mm Hg with
hydrochlorothiazide. The differences
between hydrochlorothiazide and pro-
pranolol  with respect to these
changes was significant (Pc.001) for
both the systolic and diastolic differ-
ences. During the long-term treat-
ment phase, 28 patients were termi-
nated because their diastolic BP rose
above the levels defined as terminat-
ing criteria. Twenty-five of these
patients had been randomized to pro-
pranolol, while three had been receiv-
ing hydrochlorothiazide (Pc.01).

The percentage of patients whose
diastolic BP was controlled below 90
mm Hg during the long-term treat-
ment phase was 52.8% in the pro-
pranolol group and 65.5% in the
patients receiving hydrochlorothia-
zide (P<.O3). In addition, an estimate
was made of the ability of either drug
to arrest progression of the hyperten-
sion by comparing the diastolic BP
during the prerandomization period
and the diastolic BP at the end of
posttreatment or final placebo period,
which followed the long-term treat-
ment phase. A rise of diastolic BP
during the final placebo period as
compared with the level present in
the prerandomization baseline phase
may represent possible evidence of
progression of the underlying hyper-
tension during treatment. The dia-
stolic BP was higher in the post-
treatment placebo period than in the
prerandomization phase in 32% of
the patients receiving propranolol
compared with 10% of those receiving
hydrochlorothiazide. Thus, BP in-
creased above baseline three times
more frequently after withdrawal of
propranolol treatment than after dis-
continuing hydrochlorothiazide ad-
ministration.

Race

Because there seemed to be racial
trends with regard to responses to the
two drugs, it was considered essential
to analyze the changes by race as well
as by drug. It was possible that the
greater response to hydrochlorothia-
zide detailed previously could have
been influenced by racial differences
in responsiveness. At the end of the

JAMA, Ott 22/29, 1982-Vol 248, No. 16         Propranolol and Hydrochlorothiazide, Part II-VA Study Group   2005


Table I.-Changes in BP During and After the Long-term Treatment Period, by Race'

Change

        -5.6              -19.9             -10.6             -14.7

Prerandomization

        142.1              148.6              147.8             147.0

Change

o Prerandomi.zation BP represents ??????? BP for at least two visits of that phase. Long-term treatment and posttreatment placebo phase BPS represent BP at last
vi& of each phase.

tAlso includes patients terminated in initial titration period (phase A).

Table P.-Changes in BP During Long-term Treatment Period Compared With Short-term Titration Phase

Pl_esQ of
stu<fY

               4wrsg* BP. mm Wq                      liiigrlwteenes of
                                                 DNtwAee (PI
           B                           W                          Race
   RCptMOlOl       Wydmclbicre-      Propran-    Wydrachlcre-            X
A   wydrccftterfde rt     wtls&ds     It     &I     D              WSCS   DW4   Drue

Systolic BP'         57               95              68           82
Sherbterm tttfslim    . . .      1211.1      . . .      i28.1 ,..    130.2    . .
Long-term treatment          135.8            127.7         136.8    
Change                 +7.7            +1.6     . . .     +6.6    . .
Diastolic BP         57               95              68           82
  Short-term tikation    . .       85.7      . .      86.4     . . .     85.9    . . .
Long-term treatment           88.9            87.9          91.2 ___
Ct-P           . . .      +3,2            +1.5     . . .     wi.3    

`Short-term titration phase and long-term treatment phase BPS represent BP at last visit of each phase.

128.7      . . .     . . .     . . .
130.9                
+2.2      WS    <.ool    WB


88.6      . . .     .     .
89.1                
fo.6      WB    <.oDl NS

long-term treatment phase, the num-
ber of blacks still in the trial was 152
and the number of whites was 150,
that is, the number in each group was
essentially equal. However, the racial
distribution was unequal within each
drug group (Table l), necessitating
the use of ANOVA in the analysis to
separate drug effects from race
effect.

After treatment with propranolol
the average change in systolic BP
from prerandomization baseline was
-5.6 mm Hg in blacks and -10.6 mm
Hg in whites (Table 1). With hydro-
chlorothiazide, the systolic BP change
averaged -19.9 mm Hg in blacks and
-14.7 mm Hg in whites. An analysis
of variance to attribute these differ-
ences to a pure race effect, pure drug

effect, and effect of differential racial
response within each treatment group
showed that differences between the
two drug groups and the differential
response were both highly significant.
Both racial groups without regard to
treatment exhibited a drop in systolic
BP. However, the reductions were
greater  with hydrochlorothiazide
than with propranolol in both racial
groups, although whites responded
better to propranolol than did blacks,
and the reverse with hydrochlorothia-
zide. These racial differences, how-
ever, were not significant.

A similar analysis of changes in
diastolic BP showed that hydrochlo-
rgthiazide treatment resulted in a
significantly greater lowering of dia-
stolic BP than propranolol. No racial

effect was shown, however, both
racial groups having a similar fall in
diastolic pressure without regard to
drug. Both whites and blacks re-
sponded more to hydrochlorothiazide
treatment than to propranolol treat-
ment, and the race-by-drug interac-
tion was not significant.

The mean levels of BP at the end of
the long-term treatment period were
compared with those obtained during
the short-term titration phase to
compare short-term with long-term
treatment effects by drug and race
(Table 2). Both systolic and diastolic
BP were somewhat higher in both
races and with both drugs at the end
of the long-term treatment phase.
The mean elevations with propranolol
averaged 7.6/3.2 mm Hg in blacks and

2006  JAMA, Ott 22129, 1982-Vol 248, No. 16        Propranolol and Hydrochlorothiazide, Part II-VA Study Group


Race

          Race x
Drug        DW

          .           . .
                     
          .           .
NS         <.OOl         <.OOl
. . .          . . .           .
                     
.          . .           .
NS         <.OOl          NS

.          .           .
                       
          . .           . .
NS          =.03           NS
. .                     
                     
          . . .           .
NS         <.OOl          NS

6.6/5.3 in whites. With hydrochloro-
thiazide, the mean increases were
1.50.5 mm Hg in blacks and 2.2/0.5
mm Hg in whites. The increases in BP
over time were significantly less with
hydrochlorothiazide than with pro-
pranolol (P<.OOl). Racial differences
and race-by-drug interactions with
respect to these change in BP were
not significant.

Heart Rate

There was a marked reduction in
heart rate in the group taking pro-
pranolol as compared with the pre-
randomizatiop heart rates, which av-
eraged 77.0 beats per minute before
treatment and 60.9 beats per minute
on entering the long-term treatment
period after drug titration. On the
other hand, heart rate rose slightly
after hydrochlorothiazide adminis-
tration, averaging 76.6 before treat-
ment and 79.5 after titration. The
difference between the two drugs was
highly significant (P<.OOl). These
changes  were  maintained during
long-term treatment. At the end of
the long-term treatment period, the
reduction in heart rate compared
with baseline averaged 16.2 beats per
minute with propranolol and 0.8 beats
per minute with hydrochlorothiazide.
At the end of the study after the final
two-week placebo period, heart rate
rose from propranolol treatment lev-
els to a mean value of 0.23 beats per

minute above pretreatment baseline,
while with hydrochlorothiazide the
average fell to 3.30 beats per minute
less than baseline. The difference
between the two drugs was signifi-
cant (PC.001).

Body Weight

Average body weight was greater
for the patients taking propranolol
than for those receiving hydrochloro-
thiazide. Compared with baseline, the
patients receiving propranolol gained
an average of 1.69 kg compared with a
loss averaging 1.97 kg in the group
treated  with hydrochlorothiazide
(P<.OOl). This significant difference
may have been due at least in part to
reduction in extracellular fluid vol-
ume maintained by the hydrochloro-
thiazide. There was no significant
difference in weight changes between
blacks and whites. In the final place-
bo period, the average body weight in
the propranolol group as compared
with pretreatment baseline increased
by 0.99 kg and in the hydrochlorothia-
zide group remained under the base-
line average by 0.46 kg.

Terminations

Ninety-two of 394 patients were
terminated during the long-term
treatment period, 47 for administra-
tive' reasons and 45 for medical
causes. Administrative reasons for
termination included 14 patients re-
ceiving propranolol and 18 receiv-
ing hydrochlorothiazide who stopped
treatment or failed to return, three
patients receiving each drug who
moved, two in each group who with-
drew consent, and one patient receiv-
ing propranolol and three treated
with   hydrochlorothiazide   who
stopped their drug use because of
unrelated intercurrent illnesses.

More terminations owing to medi-
cal causes occurred in the propranolol
group as compared with the patients
receiving hydrochlorothiazide. There
were 46 medical terminations, of
which 35 were associated with pro-
pranolol and 11 with hydrochlorothia-
zide (P<.OOl), that is, medical termi-
nations were approximately three
times as frequent with propranolol as
with hydrochlorothiazide. The differ-
ence was due primarily to elevation of
diastolic BP beyond acceptable limits
in patients treated with propranolol.
During the long-term treatment peri-
od, patients were terminated if their

diastolic BP was either greater than
104 mm Hg or greater than their
pretreatment diastolic BP on two suc-
cessive visits, or if the diastolic BP
was greater than 119 mm Hg on any
single visit. Among patients treated
with propranolol, 25 were terminated
because of elevated diastolic BP com-
pared with three patients in the
hydrochlorothiazide group, that is,
terminations owing to unsatisfactory
control of diastolic BP were nearly
nine times as high in the patients
treated with propranofil compared
with those receiving hydrochlorothia-
zide.

Seven of the eight terminations
related to side effects also were asso-
ciated with propranolol. There were
three patients with bronchitis and
wheezing and one patient each with
syncope, impotence, depression, or
hallucinations, compared with one
patient receiving hydrochlorothiazide
who complained of muscle cramping.
Myocardial infarction was diagnosed
or suspected in two patients receiving
hydrochlorothiazide and two receiv-
ing propranolol. Stroke occurred in
three patients taking hydrochlorothi-
azide. Cancer developed in three
patients all receiving hydrochlorothi-
azide. The incidence of these morbid
events was too low to draw any
conclusions as to possible drug influ-
ences.

Adverse Reactions

All of the patient's complaints were
reported, and those that were not
present during the prerandomization
period were considered as possible
side effects of the drug used. The
results were analyzed by calculating
the fraction of visits for which a given
new complaint was either volunteered
or elicited. Some of the complaints
were noted in a significantly higher
frequency with one drug than with
the other. However, because of the
large variety of complaints it was
probable that some correlated "sig-
nificantly" by chance alone.

The various complaints are listed in
Table 3. The following complaints
were associated significantly with the
patients treated with hydrochlorothi-
azide as compared with those taking
propranolol: tachycardia, diarrhea,
constipation, impotence, tinnitus, dry
mouth, lumbar pain, loss of libido,
edema, and numbness and tingling.
The complaints associated signifi-

JAMA, Ott 22129, 1982-Vol 248, No. 16.        Propranolol and Hydrochlorothiazide. Part II-VA Study Group   2007


cantly more with propranolol than
with hydrochlorothiazide were indi-
gestion, insomnia, vivid dreams, de-
pression, hallucinations, blurred vi-
sion, and swelling of the hands. Many
of these complaints had a low inci-
dence and some were bizarre, such as
edema significantly associated with
the diuretic. The most frequent com-
plaints that were noted in 1% or more
of patient visits were insomnia, swell-
ing of the hands, and vivid dreams,
which were associated significantly
with propranolol, whereas diarrhea,
impotence, constipation, and numb-
ness were associated with hydrochlo-
rothiazide. All of these complaints
were relatively infrequent, however,
none being noted in more than 3% of
patient visits.

Blood Chemistries

The changes in blood chemistries
from the pretreatment levels are
shown in Table 4. Significant changes
in mean values after propranolol
administration included a slight rise
in serum potassium concentration,
fasting blood glucose level, and cal-
cium level. Serum cholesterol levels
averaged 4% higher than control and
triglyceride values rose 25% above
the baseline level. During long-term
treatment with hydrochlorothiazide
the average uric acid concentration
increased 21% as compared with the
pretreatment value, the serum potas-
sium level decreased 13%, and the
serum urea nitrogen level increased
17%. All of these changes were signif-
icant. Fasting glucose values in-
creased slightly, but the change was
not significant. Serum cholesterol and
triglyceride levels surprisingly de-
creased slightly but not significantly.

Compared with short-term treat-
ment (Table 5), the changes in blood
chemistry findings occurring with the
passage of time were generally small
and not clinically important. With
propranolol, the average changes at
the end of long-term treatment as
compared with short-term treatment
were a decrease of 6% in serum urea
nitrogen levels and of 2% in potas-
sium levels. Serum triglyceride levels
rose further, the increase averaging
6%. With hydrochlorothiazide, there
was no further change during long-
term treatment except for slight
increase in potassium values and fall
in creatinine levels as compared with
the early phase of treatment. Serum

Table 3.-Complaints Not Present During Prerandomization Period
          Present During Treatment o

?????????
Swelling of hands
Vivid dreams
Depression
l-t&JdWtbllS
Indigestion
F5klm4utslo"
Hydrcchlorathiaride-associated
Mamhea
Impotence
consflpation
Numbness
LUFnbsr pat"
Loss of libido
Edema
Tinnitus
Numbneea, tln@ttn@

??????
0.0260
0Slt66
0.0072
imQ20
0.0013
moo%


0.0152
0.0203
w?o%e
0.0064
0.0034
0.0003
cm00o
0.0000
o.m5

0.010%
0.0103
0.0086
0.0040
o.cm2
0.0000
a.o00o


a.0970
0.029 1
0.0184
0.0150
Q.W7
0.0065
0.0066
0.0055
0.0031

`Complaints present during treatment significantly mere often with one drug than with the other.

triglyceride levels decreased 17%, but
the change was not significant
(P=.O7).

During long-term treatment, no
patients in the group treated with
propranolol exhibited a serum potas-
sium level below 3.5 mEq/L. In the
patients treated with hydrochlorothi-
azide, 41% exhibited levels below 3.5
mEq/L, of whom 6% were below 3.0
mEq/L. In both treatment periods,
uric acid levels of 10 mg/dL developed
in approximately 12% of the patients
treated with hydrochlorothiazide, as
opposed to none in the group treated
with propranolol. Fasting blood glu-
cose levels of 150 mg/dL or above
occurred in 6% of the patients receiv-
ing hydrochlorothiazide and in 4% of
those receiving propranolol. These
percentages represent an increase
above pretreatment baseline of 3%
for the patients treated with hydro-
chlorothiazide and no change for the
propranolol group. There was no sig-
nificant change from pretreatment
baseline in the percentage of patients
with cholesterol levels of 300 mg/dL
or higher in the patients treated with
either propranolol or hydrochlorothi-
azide. The percentage of patients
exhibiting elevated triglyceride levels
of 300 to 399 mg/dL rose from 4.2% in
the pretreatment period to 8.6% in
the titration phase to 11.7% in the
long-term treatment period. The per-

cent increases with hydrochlorothia-
zide were from 3.8% during the pre-
treatment phase to 5.5% and 5.8% in
the titration and long-term treatment
period, respectively.

Dosage Requirements

After the initial titration of dosage
and during the long-term treatment
period, 37.6% of the patients treated
with propranolol and 20.9% of pa-
tients receiving hydrochlorothiazide
required further increases of dosage
(Pc.015). By contrast, only 0.8% of
patients receiving propranolol and
4.0% of those taking hydrochlorothia-
zide had their doses reduced, the total
number of these being too small to
achieve significance. These differ-
ences probably reflected the lower
percentage of patients controlled with
propranolol as compared with hydro-
chlorothiazide.

COMMENT

As described in the preceding arti-
cle, the baseline characteristics of the
patients assigned to each regimen
were similar. Because of fewer pa-
tients attaining goal BP with pro-
pranolol as compared with hydro-
chlorothiazide, more patients taking
propranolol were terminated in the
initial titration phase than was the
case with hydrochlorothiazide. As a
result, at the beginning of the long-

2008  JAMA, Ott 22129, 1982-Vol 248, No. 16         Propranolol and Hydrochlorothiazide, Part II-VA Study Group


                          Table 4.-Changes in Blood Chemistries*                        ,


                      Propranolol Hydroohlarklr               Hydmchlor~&o

                        Protreatment     Mean                PF&WMlVtUSt    Mban           llkbtwbbn
     Blood Toat        n       Meen     ~miw    P ' n       wbon     @`=W'     P    ma $5)
  Uric acid, mg/dL        122       6.56       0.10      NS     174       6.47        1.37     c.001     <.OOl
Potasstum. mEq/L       121       4.23       Q. 17     <.ool    174       4.20      -0.6%    c.uclQ     <.itol
  Creatinine, mgldL       123       1.16       0.01      NS     173        1.16       0.03      NS       NS
urea nitro@sn. mg/dL     120       14.1        0.46      NS     175       14.3        2.6     <.OOt     <.tm
  Fasting glucose, mg/dL    119       99.6        6.4     <.OOl    174      101.7        4.7     <.05       NS
Cholesteml, mg/dL       121      222.3        8.7      =;.02     167      22%,2       -3.0      NS       NS
  Triglycerides, mg I dL      116      166.9       42.2     <.Ol     170      166.7       -3.3      NS      <.05

1  calclum,moldl         131       3.27                           '
                                     CMd     1.001    loo       $.4&I       -0.13    -aI6      <.6oi
`During long-terni treatment period compared with pretreatment period.

Table L-Changes in Blood Chemistries*

   efoad Tea         n
Uric acid, mg/dL         107
Potassium. &q/L        106
Creatinine, mgldL        107
Urea nitrogen. mg /dL      104
Fasting glucose, mg/dL     103
Chotesterol, mg/dL        103
Triglycerides, mg / dL       102
Calcbm. ms/dL          121

Proprbnolot NydrocMaride

srturt-~m     Meen
Wem     Change
  6.71       0.0
  4.49      -0.10
  1.19      -0.01
  16.32      -0.8B
103.6        3.13
217.0        0.6
197.7       11.7
  9.64      -0.03

P      n
NS      152
<.a2    tr64
NS      154
<.02     166
NS      153
MS      147
NS      147
NS      89

HydfothiaroPiazlde

SlIort-tortn     Moan
atban     -w
  6.05       -0.15
  3.50       a.03
  1.24       -0.04
  17.02      -0.28
106.0        0.60
231.0       -7.7
219.6       -37.6
  0.57      -0.02

P
NS
N%
<.03
MS
NS
<.a?
=.07
NS

DltWrenob
sowom
@fwMwl
  NS
<St2
  NS
  MS
  NS
  NS
  NS
  NS

*During long-term treatment period compared with short-term treatment period

term treatment phase the number of
patients receiving each drug was
slightly different, with 182 patients
receiving propranolol and 212 taking
hydrochlorothiazide. Since the differ-
ence was due to the selective loss of
propranolol nonresponders, the effect
should be to influence the results in
the long-term treatment period in
favor of propranolol as compared
with hydrochlorothiazide. However, it
seems. likely that this influence, if
any, was small and probably did not
significantly affect the results.

By the end of the long-term treat-
ment period an additional greater
number of patients were terminated
in the group receiving propranolol.
The greater number of terminations
was due primarily to the lesser effec-
tiveness of propranolol to control the
diastolic BP within acceptable limits.
Therefore, by the time of completion
of the long-term treatment period,
there were 29% more patients re-
maining in the hydrochlorothiazide
group than in the group receiving
propranolol. However, the BP differ-
ences between the two regimens still
were large despite elimination of the
patients unresponsive to propranolol
and probably could not be ascribed to
the differences in sample size result-

ing from attrition. The differences
would have been even greater if the
BP of the terminated patients also
had been included.

The greater antihypertensive effec-
tiveness of hydrochlorothiazide as
compared with propranolol was evi-
denced by several criteria. The reduc-
tions of BP, both systolic and diastol-
ic, were significantly greater with
hydrochlorothiazide. The difference
was most marked with respect to
systolic BP (P<.OOl). The difference
in the degree of fall in diastolic BP
between the two drugs was not great
(P=.O3). This was partly due to the
fact that propranolol was associated
with a greater reduction of diastolic
than of systolic BP. This effect of
propranolol may be due at least in
part to the slowing of heart rate,
which, by lengthening the diastolic
runoff period, contributes to the low-
ering of diastolic BP. After hydro-
chlorothiazide treatment, however,
the systolic fall was greater than the
diastolic.

Evidence of the more sustained
effectiveness of hydrochlorothiazide
is provided by the comparison with
the BP at the end of the initial
titration period. Compared with this
short-term treatment period, both

systolic and diastolic BP rose signifi-
cantly higher during the long-term
treatment period in more patients
receiving propranolol than in those
receiving hydrochlorothiazide. Fur-
ther evidence of the lesser effective-
ness of propranolol is provided by the
cases terminated because of increased
diastolic BP. Eighty-six percent of the
terminations due to this cause had
been receiving propranolol.

The dosage requirement also was
less with hydrochlorothiazide than
with propranolol. Three times more
patients responded to a low dose of
hydrochlorothiazide (25 mg twice dai-
ly) as responded to a low dose of
propranolol (40 mg twice daily),
although the percentage of patients
receiving the highest doses used in
the titration procedure was essential-
ly the same with the two drugs.
However, the number of patients
requiring increased doses during the
long-term treatment phase was near-
ly twice as high in the propranolol
group of patients as in those receiving
hydrochlorothiazide.

Additional evidence of the lesser
antihypertensive effectiveness of pro-
pranolol as compared with hydrochlo-
rothiazide is shown by the percentage
of patients whose diastolic BP was

JAMA, Ott 22129, 1982-Vol 248, No. 16

Propranolol and Hydrochlorothiazide, Part II-VA Study Group   2009


controlled below 90 mm Hg. This goal
was achieved in 65.5% with hydro-
chlorothiazide and in 52.8% with pro-
pranolol, a significant difference. In-
clusion of the 26 losses among
patients receiving propranolol 2, three
treated with hydrochlorothiazide be-
cause of the development of increased
BP would make the difference even
greater. Finally, after discontinuation
of treatment with both drugs at the
end of trial, diastolic BP rose above
the pretreatment placebo level three
times more frequently with proprano-
101 than with hydrochlorothiazide.
Therefore, by several different in-
dices of effeciiveness, propranolol
alone was less active as an antihyper-
tensive agent than was hydrochloro-
thiazide alone. The mean reduction of
BP, both systolic and diastolic, was
significantly less; significantly fewer
patients had their diastolic BP con-
trolled; considerably more patients
had to be removed because of
increased diastolic BP in the propran-
0101 group than in the patients receiv-
ing hydrochlorothiazide; fewer pa-
tients were controlled on low doses of
propranolol than on the lower dose of
hydrochlorothiazide; almost twice as
many patients in the propranolol
group as compared with hydrochloro-
thiazide required increases in dosage
during the long-term treatment peri-
od, and after discontinuation of treat-
ment the number of patients whose
diastolic BP rose above the pretreat-
ment baseline level was three times
greater in the patients taking pro-
pranolol than was the case in the
group receiving hydrochlorothiazide.

Other investigators have not found
such clear differences in the antihy-
pertensive effectiveness of diuretic as
compared with P-blockers. Berglund
and Andersson,5 in a randomized trial
of six years' duration, found no signif-
icant difference in BP response
between 40 and 80 mg per day of
propranolol hydrochloride and 2.5 to
5.0 mg per day of bendroflumethia-
zide. Paterson and Dollery" carried
out a crossover trial in 11 patients
with mild hypertension in which they
compared 80 and 240 mg per day of
propranolol with hydrochlorothiazide
50 mg daily. The BP was consistently
lower with the diuretic, but probably
because of the small sample size the
difference was not significant. Also,
the daily dose of hydrochlorothiazide
was only one fourth the maximum

dose used in the present study. See-
dat' gave atenolol, 100 mg daily, in a
crossover trial with chlorthalidone, 25
mg daily, to 24 black South Africans.
Atenolol had no more effect than
placebo on BP, while chlorthalidone
was associated with a small but not
significant reduction. The reason that
the reduction was small with the
diuretic may have been due to the
rather small dose of chlorthalidone
used in his study. The relatively poor
effect of the &blockers in black
patients is consistent with our experi-
ence.

Few cardiovascular complications
and no deaths related to cardiovascu-
lar disease occurred during the long-
term treatment period. In a prior
Veterans Administration trial4 com-
paring various regimens, in the treat-
ment of mild hypertension, including
propranolol plus hydrochlorothiazide
and reserpine plus hydrochlorothia-
zide, there were two cardiovascular
deaths associated with propranolol
plus hydrochlorothiazide, compared
with none with reserpine plus hydro-
chlorothiazide. In the six-year trial of
Berglund and Andersson on propran-
0101 `u thiazide diuretic, there were
three deaths associated with propran-
0101 as compared with one with the
diuretic. Although the number of
events is too small to draw any defi-
nite conclusions, these studies do not
support the opinion that propranolol
is more protective against cardiovas-
cular deaths than is thiazide. How-
ever, there is evidence that /3-adrener-
gic blocking agents are effective in
the prevention of sudden death fol-
lowing recovery from myocardial in-
farction.*.9

Side effects sufficiently serious to
terminate the patient from the trial
occurred in a ratio of 7 to 2, proprano-
101 compared with hydrochlorothia-
zide. Certain biochemical side effects,
particularly hypokalemia, were con-
siderably higher with hydrochlorothi-
azide than with ' propranolol. The
unusually high incidence of hypokale-
mia may have been related to the dose
of hydrochlorothiazide, which ranged
between 50 and 200 mg per day.
Nonterminating complaints were
somewhat more numerous in patients
treated  with hydrochlorothiazide
than in the group receiving proprano-
101. Their importance and their rela-
tionship to the drug, however, is
questionable. For example, impotence

and loss of libido together was com-
plained of in 0.0203 of all patient
visits with propranolol and 0.0350 in
those taking hydrochlorothiazide. The
difference,  which was significant,
involved only 1.5% more patient vis-
its in the hydrochlorothiazide than in
the propranolol group. Considering
the difficulty in obtaining reliable
information from the patient and
with the increased opportunity for
chance association when a wide varie-
ty of complaints are being itemized,
the nonterminating side effects data
is probably the least reliable of any of
the results reported herein.

The British Medical Research
Council trial"' has reported on adverse
reactions  to propranolol and to
diuretic in approximately 10,000 pa-
tients with mild to moderate hyper-
tension. Impotence was complained of
in 16.2% of men receiving the diuret-
ic, 13.8% receiving propranolol, and
8.9% taking placebo. However, ap-
proximately three times as many
patients were terminated because of
impotence in the diuretic group as
compared with patients taking pro-
pranolol. This is considerably higher
than was observed in the present
trial. In the study by Berglund and
Anderssons  subjective side effects
were equally distributed between pro-
pranolol and diuretic. The incidence
of gout and diabetes mellitus also
were the same in both groups. With
the exception of the British trial, it
seems that both drugs were well
tolerated,  with slightly but not
markedly more complaints of impo-
tence with hydrochlorothiazide than
with propranolol.

The data suggested but did not
conclusively demonstrate a racial dif-
ference in responsiveness to the two
drugs. The mean reduction in BP with
propranolol was greater in whites
than in blacks, while with hydrochlo-
rothiazide the greatest mean reduc-
tion was greater in blacks. However,
the reductions in diastolic BP with
hydrochlorothiazide were greater
than with propranolol in either race,
and the differences between the drugs
by race were not significant and could
not account for the lesser effect of
propranolol as compared with hydro-
chlorothiazide. Seedat' also found
that a diuretic was more effective
than a ,&blocker in blacks. Hyperten-
sive blacks tend to have higher plas-
ma volumes and lower plasma renin

2010 JAMA, Ott 22129, 1982-Vol 248, No. 16          Propranolol and Hydrochlorothiazide, Part II-VA Study Group


activity than white hypertensive per-
sons."-13 According to Laragh,' such
"low renin" or "volume dependent"
hypertension will respond better to
diuretics than to fl-adrenergic block-
ing drugs, which was actually the
case in both Seedat's' and the present
study. However, while renin profiling
may be informative in revealing gen-
eral trends, most investigators have
not found it to be a reliable guide for
treating individual patients. Renin
profiling"," was carried out in the
present trial, and the results will be
reported in another communication.
Other possible mechanisms for the
black-white difference in responsive-
ness to hydrochlorothiazide and pro-
pranolol have been discussed in the
preceding article. One of the most
striking racial differences found in
the present study was the presence of
a considerably reduced potassium
excretion,16 probably reflecting a di-
minished dietary intake in blacks as
compared with whites.

The changes in BP secondary to
treatment were similar during the
acute titration phase and the long-
term phase. The decrements of systol-
ic and diastolic BP were essentially of
the same degree acutely and over the
long term. The fall of BP was greater
with hydrochlorothiazides than with
propranolol at both times, and the
effectiveness of hydrochlorothiazide
seemed to be greater in blacks than in
whites both acutely and in the long
term.

After propranolol treatment, serum
triglyceride levels increased by 25%

1. Laragh JH: Vasoconstriction-volume analy-
sis for understanding and treating hypertension:
The use of renin and aldosterone profiles. Am J
Med 1973;55:261-274.
2. Woods JW, Pittman AW, Pulliam CC, et al:
Renin profiling in hypertension and its use in
treatment with propranolol and chlorthalidone.
N Engl J Med 1976;294:1137-1143.
3. Holland OB, Nixon JV, Kuhnert LaV:
Diuretic-induced ventricular ectopic activity.
Am J Med 1981;70:762-768.
4. Propranolol in the treatment of essential
hypertension, Veterans Administration Cooper-
ative Study Group on Antihypertensive Agents.
JAMA 1977;237:2303-2310.
5. Berglund G, Andersson 0: Beta blockers or
diuretics in hypertension? A six year follow-up
of blood pressure and metabolic side effects.
Laneet 1981;1:744-`747.
6. Paterson JW, Dollery CT: Effect of pro-
pranolol in mild hypertension. Lancet 1966;
21148.1150.
7. Seedat YK: Trial of atenolol and ehlorthali-
done for hypertension in black South Africans.
Br Med J 1980;281:1241-1243.
8. Timolol-induced reduction in mortality and
reinfaretion in patients surviving acute myocar-

in the present study. This is similar to
the increases reported by other inves-
tigators."." Propranolol also has been
reported to reduce serum high-densi-
ty-lipoprotein cholesterol levels, but
to produce no significant change in
total cholesterol levels.`* We also
observed little change in total serum
cholesterol levels in the present
study. Our results, however, are at
variance with other investigators
with respect to the lipidemic effects
of thiazide diuretics. While no signifi-
cant changes were found in serum
cholesterol or triglyceride concentra-
tions after hydrochlorothiazide treat-
ment, most others have observed
increases in both.`9-2' The reason for
the discrepancy is not clear. System-
atic technical errors did not seem to
be involved because the determina-
tions were carried out separately in
each hospital using an autoanalyzer
method. The European Working Par-
ty on High Blood Pressure in the
Elderly also did not find an increase
in total serum cholesterol levels dur-
ing long-term treatment with hydro-
chlorothiazide in combination with
triamterene." Thus, the increase in
total cholesterol does not seem to be a
consistent accompaniment of thiazide
treatment.

Fewer patients responded to pro-
pranolol than to hydrochlorothiazide
despite the fact that the dosage was
titrated to higher levels in the group
receiving propranolol. Responders to
the lower dose range of propranolol
were less than was the case for the
lower dose range of hydrochlorothia-

References

dial infarction, -the Norwegian Multicenter
Study Group. N Engl J Med 1981;304:801-807.
9. P-Blocker Heart Attack Study Group: The
B-blocker heart attack trial. JAMA 1981;
246:2073-2074.
10. Adverse reactions to bendroflumethiazide
and propranolol for the treatment of mild hyper-
tension, Medical Research Council Working Par-
ty on Mild to Moderate Hypertension. Lancet
1981;23539-543.
11. Lillez JL, Hsu L, Stone RA: Racial dispari-
ty of plasma volume in hypertensive man. Ann
Intern Med 1976;&1:707.
12. Messerli FH, DeCarvalho JGR, Christie B,
et al: Essential hypertension in black and white
subjects: Hemodynamie findings and fluid vol-
ume state. Am J Med 1979;67:27-31.
13. Mitas JA II, Holle R, Levy SB, et al: Racial
analysis of the volume-renin relationship in
human hypertension. Arch Intern Med 1979;
139~157-160.
14. Brunner HR, Gavras H: Clinical implica-
tions of renin in the hypertensive patients.
JAMA 1975;233:1091-1093.
15. Woods JW, Pittman AW, Pulliam CC, et
al: Renin profiling in hypertension and its use in
treatment with propranolol and chlorthalidone.

zide. Combined with the greater cost
of propranolol, it is apparent that the
expense of treating most patients
with propranolol considerably ex-
ceeds that of using hydrochlorothia-
zide.

This study was supported by a grant from
Ayerst Laboratories, Inc.
The following persons participated in the
study:

Co&airmen: Edward D. Freis, MD (Washing-
ton, DC) and Barry J. Materson, MD, (Miami).
Priicipl Investigators: Frederick N. Talmers,
MD (Allen Park, Mich); William C. Cushman,
MD (Jackson, Miss); Harold Sehnaper, MD (Bir-
mingham, Ala); Thomas J. White, MD, (Mem-
phis); Khin Mae Hla, MD, and Orlando Fernan-
dez, MD (Miami); Eli A. Ramirez, MD (San Juan,
Puerto Rico); Ibrahim Khatri, MD (Washington,
DC).

Nurses: Barbara Gregory, RN, and Madeline
Metcalfe, RN (Washington, DC); Julie Pawelak,
RN (Allen Park, Mieh); Pauline Derrington, RN
(Jackson, Miss); Susan Reeee, RN, and Kristina
Grossman, RN (Memphis); Mary H. Smith, RN,
and Eileen Haran, RN (Miami); Maria Natal, RN
(San Juan. Puerto Rico); Donald Quinn, PA

iBirmingham, Ala).
Biostatisticians: Bor Ming Ou, MS, Samuel B.
Lindle, PhD, Domenie Reda, MS.
                Flamenbaum,

Special Renin Laboratory: Walter
MD, and Robert Hamhurger, MD.
            _. . .

Central Research hmrmaclst: Larry Young,

RPh.

Operations Committee: Gilbert MeMahon, MD,
Ray Gifford, MD, and C. Morton Hawkins,

PhD.

Consultants: James R. Oster, MD, Ezra Lamdin,
MD (Averst Laboratories), Shig Ochi, PhD, and

J. R.`Thomas, MD.

Hines Cooperative Studies Program Coordinating
Center Human Rights Committee: Jennie McKay;
Patrick Moran; Mary Davidson, PhD; Kenneth
Elmer; Rev Martin Feldbush.
Coopentive Studies Program Central Administra-
tion: James A. Hagans, MD, PhD, and Ping
Huang, PhD (VA Central Office, Washington,
DC); Kenneth James, PhD, and William G.
Henderson, PhD (Cooperative Studies Program
Coordinating Center, Hines, Ill); Mike Sather,
RPh, MS (Cooperative Studies Program Re-
search Pharmacy Coordinating Center, Albu-
querque).

N Engl J Med 1976;294:1137-1143.
16. Urinary Na'/K+ ratio is much higher in
blacks and correlates with diastolic blood pres-
sure in hypertensive patients, Veterans Admin-
istration Cooperative Studies Group on Antihy-
pertensive Agents, abstracted. Cardicwasculcw
Disease Epidaniology Newsletter, Council on
Epidemiolom. American Heart Association
1982;31:37. --.

17. Shaw J, England JDF, Hua ASP: Beta-
blockers and plasma triglycerides. Br Med J
1978;1:968.

18. Leren P, Helgeland A, Holma I, et al:
Effect of propranolol and prazosin on blood
lipids: The Oslo study. Lamet 1980,2:4-6.
19. Ames RP, Hill P: Increase in serum lipids
during treatment of hypertension with chloro-
thiazide. Zmxet 1976;1:721-725.
20. Schnaper H, Fitz A, Frohlich E, et al:
Chlorthalidone and serum cholesterol. Luncet
1977;2:295. .

21. Smith WMeF: Diuretics and eholestero
elevation. JAMA 1979;242:1612.
22. Amery A, Bulpitt C, de Sehaepdryver A:
Thiazides and serum cholesterol. Lancet 1980;
2473.

JAMA, Ott 22129, 1982-Vol 248, No. 16

Propranolol and Hydrochlorothiazide, Part II-VA Study Group   2011

Punted and Published in the United States of America